SGLang 重构 CUDA Graph 支持，Breakable CUDA Graph 成 prefill 默认方案
Blog Advanced CUDA Graph Techniques in Inference CUDA Graphs promise to remove kernel-launch overhead, but getting close to that benefit in a real inference engine requires graphing as much of the workload as possible without sacrificing compatibili... SGLang Team
AI 导读SGLang 重构 CUDA Graph 支持，通过 runner/backend 接口拆分使不同捕获策略可复用。其社区首创的 Breakable CUDA Graph（BCG）现为 prefill 默认方案，代码量仅为 torch.compile 方案的约四分之一（521 行对比 1,771 行），构建速度快 3.8–5.2 倍。
来源：LMSYS Blog · lmsys.org